Clean Energy Development and Public Health: An Empirical Analysis of Cross-Regional Gas Transmission Infrastructure in China

Abstract

Promoting the clean energy transition is crucial for environmental sustainability and public health. Utilizing data from the China Health and Nutrition Survey (CHNS) spanning 2006–2015, this study employs a Difference-in-Differences (DID) model, treating China’s West–East Gas Pipeline Project (WEGT) as a quasi-natural experiment to evaluate the causal impact of natural gas infrastructure expansion on residents’ health. The empirical results indicate that the WEGT significantly improved public health, reducing the probability of self-reported recent illness by approximately 8.2 percentage points. Heterogeneity analysis shows more pronounced effects among urban residents and the elderly. Mechanism analysis reveals that the infrastructure improves health primarily by optimizing household energy structures and reducing industrial pollution emissions. Furthermore, the “Coal-to-Gas” policy synergistically enhances these health benefits. Economic co-benefits analysis estimates that the project reduced individual annual medical expenditures by approximately 540 RMB and increased the probability of employment by 6.9%. These findings provide empirical evidence for deepening supply-side structural reforms in energy and support the realization of the United Nations Sustainable Development Goals (SDGs), specifically by demonstrating how resilient infrastructure (SDG 9) enables affordable clean energy (SDG 7), which in turn delivers good health and well-being (SDG 3).

1. Introduction

Health is a fundamental prerequisite for human development and a primary goal of socioeconomic progress. Governments worldwide prioritize public health, and the United Nations lists health as a core component of the Human Development Index and the Sustainable Development Goals (SDGs). In China, the government has strategically prioritized public health through initiatives such as the “Healthy China 2030” plan. By 2022, China’s average life expectancy had risen to 77.93 years, marking significant progress in national health outcomes. This improvement depends not only on the continuous perfection of medical security [1] but also on the sustained improvement of environmental quality. Consequently, the development of clean energy has emerged as a critical pathway to reduce pollution and improve ecological conditions, helping to establish a long-term mechanism for low-carbon energy transition and fostering endogenous drivers for green growth.

Energy serves as the foundation and engine of economic development and is a crucial aspect of ecological civilization construction. Given China’s resource endowment characterized by coal dominance, there is an urgent need to promote clean energy development, which has triggered extensive academic debate regarding the paths and effects of energy transition [2]. On the one hand, regarding implementation paths, existing literature on energy transition primarily focuses on environmental governance [3], energy efficiency [4], financial development [5], and energy technology [6]. However, discussions specifically addressing mega-energy infrastructure projects, such as the West–East Gas Pipeline Project (WEGT), remain relatively scarce. On the other hand, regarding transition effects, existing studies mainly focus on the macro-level, examining impacts on the economy and environment [7]. Although recent scholarship has begun to utilize micro-survey data to evaluate the health benefits of specific clean heating policies [8], empirical evidence concerning large-scale energy infrastructure projects, such as the WEGT, remains scarce. Few studies have explored the positive externalities generated at the micro-individual level. This gap exists partly due to endogeneity issues, such as omitted variable bias or individual preferences, which complicate the causal inference between energy transition and individual status. The WEGT, representing an exogenous shock, provides an ideal opportunity to utilize a quasi-natural experiment framework to identify the positive health externalities of clean energy development rigorously.

To address the structural imbalance between clean energy distribution and demand in China, the State Council approved the launch of the West–East Gas Pipeline Project (WEGT) in February 2000, with full-scale construction commencing in July 2002. According to statistics from the National Energy Administration, by the end of 2022, the WEGT had cumulatively transmitted over 800 billion cubic meters of natural gas. This effectively replaced 1.07 billion tons of standard coal and reduced emissions of carbon dioxide and dust by 1.171 billion tons and 581 million tons, respectively [9]. It is important to qualify that while natural gas is technically a fossil fuel, it is widely recognized as a “bridge fuel” or “cleaner energy” within the context of China’s coal-dominated energy system. Compared to coal combustion, natural gas generates negligible sulfur dioxide and particulate matter, along with significantly lower carbon emissions. Therefore, throughout this study, we use the term “clean energy” to denote this relative environmental improvement over traditional solid fuels, rather than equating it with zero-emission renewable energy sources. Consequently, the share of natural gas in China’s primary energy consumption rose from 2.4% in 2003 to 8.4% by the end of 2021. This infrastructure has not only provided energy security for high-quality development but also injected clean momentum into achieving China’s “carbon peaking and carbon neutrality” goals and advancing ecological civilization initiatives. Evidently, the WEGT has generated immense environmental benefits; however, the associated health externalities remain to be rigorously examined.

The development of clean energy impacts resident health through multiple dimensions. On the one hand, it drives the transition of household energy structures (Direct Impact). When households rely on firewood and coal as primary fuels, the inefficient combustion of these fuels produces substantial indoor air pollutants. The WEGT facilitates the substitution of polluting fuels with clean energy in households, thereby optimizing the energy structure, reducing indoor pollution, and directly improving resident health. On the other hand, it promotes industrial pollution reduction (Indirect Impact). Extensive literature has established that air pollution has a significant impact on public health [10]. Industrial production is a major source of such pollution. Following the operation of the WEGT, the reduction in pollutants from industrial energy consumption has led to sustained improvements in regional air quality, thereby indirectly enhancing public health levels.

In this study, focusing on the WEGT Line 2, we utilize data from the China Health and Nutrition Survey (CHNS) and a Difference-in-Differences (DID) model to empirically assess the project’s impact on resident health. The results indicate that, as a landmark project of the Western Development Strategy, the WEGT significantly reduced the probability of illness among residents in regions along the Pipeline. This conclusion remains robust after excluding interference from concurrent energy and environmental policies, conducting placebo tests, and performing a series of sensitivity analyses. Mechanism analysis reveals that the increased supply of natural gas promotes the transformation of household energy structures. Simultaneously, enterprises utilize natural gas as a substitute fuel, which helps reduce pollution emissions. The improvement in city-level environmental quality further validates this micro-level impact channel. Further analysis indicates heterogeneity in these effects, with more pronounced health benefits observed in urban areas with larger-scale natural gas usage and among the elderly population, who are more sensitive to air pollution. Additionally, environmental regulations such as the “Coal-to-Gas” policy synergistically enhance the health benefits of the WEGT. Welfare analysis demonstrates that the project has reduced individual medical expenses and regional healthcare expenditures while promoting local employment.

This study contributes to the existing literature in three main aspects:

First, it enriches the evaluation of the implementation effects of the WEGT. Previous studies have primarily approached this topic from macro- or enterprise-level perspectives [11], paying relatively less attention to its impact on micro-individuals. This paper focuses directly on resident health from a micro-individual perspective. Furthermore, it explores the welfare effects in terms of reduced medical costs and increased employment levels following health improvements, providing empirical evidence for the role of the WEGT in promoting the “Healthy China” strategy.

Second, it expands the research on determinants of resident health. While existing literature has examined the impact of environmental issues on health and the environmental effects of the WEGT [12], as well as its effect on urban carrying capacity [13], few studies have focused on the downstream effects of environmental quality improvements. By placing the analysis of resident health against the backdrop of China’s rapid clean energy development and using the WEGT as an entry point, this study systematically explores the significant impact of clean energy development on enhancing public health levels.

Third, it supplements the logical mechanisms through which clean energy development improves resident health. This paper first explores how the WEGT improves resident health by inducing behavioral changes in micro-energy consumers—specifically through optimizing household energy structures and reducing corporate pollution emissions. It then provides supplementary verification through macro-level improvements in urban environmental performance. These findings offer theoretical support for government initiatives to promote resident health through clean energy development and validate the significant historical significance and practical value of the WEGT in synergistically advancing the energy revolution and the “Healthy China” strategy.

2. Context and Literature Review

2.1. Institutional Background: The West–East Gas Pipeline Project

Since China’s reform and opening-up, it has experienced surging energy demand. However, the nation’s resource endowment, historically characterized by coal dominance, has led to an irrational energy supply structure. As of the end of 2022, coal still accounted for 56.2% of total energy consumption [14]. Excessive coal consumption has resulted in deteriorating air quality and a series of environmental and health challenges. Consequently, promoting the structural optimization of energy supply and accelerating the development of clean energy have become critical pathways to resolve the conflict between economic growth and environmental sustainability.

In this context, China has actively explored various alternatives to coal, including hydropower, wind, and nuclear energy. Among these, natural gas—as a high-quality, efficient, and relatively clean fossil fuel—has emerged as a vital substitute. However, a significant spatial mismatch exists in China’s energy landscape: natural gas reserves are predominantly located in the western regions, while energy consumption is concentrated in the economically developed eastern coastal areas. To address this imbalance and ensure energy security, the State Council approved the launch of the West–East Gas Pipeline Project (WEGT) in 2000.

The WEGT is a massive infrastructure network consisting of four phases. While Line 1 commenced operations in 2004, this study specifically focuses on the Second West–East Gas Pipeline (Line 2), which represents the first strategic channel introducing overseas natural gas (primarily from Central Asia) into China. Construction of Line 2 began in February 2008, and it became fully operational in December 2012. The Pipeline spans 9102 km, starting from Horgos in Xinjiang, extending south to Guangzhou and Hong Kong, and east to Shanghai. It traverses 14 provinces and municipalities, including Xinjiang, Gansu, Ningxia, Shaanxi, Henan, Hubei, Jiangxi, Hunan, Guangdong, Guangxi, Zhejiang, Anhui, Jiangsu, and Shanghai. Line 2 is characterized by its extensive coverage and large beneficiary population, making it an ideal quasi-natural experimental setting for analyzing regional health disparities.

The expansion of the WEGT has significantly increased the share of natural gas in China’s primary energy mix. National natural gas consumption rose from 147.1 billion cubic meters in 2012 (4.8% of total energy consumption) to 364.6 billion cubic meters in 2022 (8.4% of total consumption). By connecting the energy-rich West with the demand-heavy East, the WEGT not only optimizes the regional energy structure but also facilitates the “green” low-carbon transition of the economy [15]. The substantial environmental benefits of this substitution—specifically the reduction in sulfur dioxide and particulate matter emissions—provide the theoretical basis for the health effects examined in this paper.

2.2. Literature Review

2.2.1. Energy Transition: Drivers and Effects

Energy transition refers to the structural transformation of the energy supply side [16]. Since the late 1970s, China has relied on low-cost and abundant coal resources, forming an extensive economic growth model characterized by high inputs, high output, and high pollution. With the shift toward sustainable economic development, China is transitioning from high-speed growth to high-quality growth, necessitating corresponding adjustments and optimization of the energy supply structure. Existing literature on energy transition focuses primarily on transition pathways and their effects. Regarding pathways, studies have extensively discussed driving mechanisms, including policy interventions [17], financial development and income growth [18], energy price changes and subsidies [19], and public environmental awareness [20]. However, few studies have examined energy transitions driven by mega-infrastructure projects like the WEGT. Regarding effects, current research largely emphasizes economic growth [21], environmental quality improvement [22], and welfare effects [23], often overlooking the positive externalities related to public health. Based on this, this paper focuses on the WEGT—a mega-trans-regional energy allocation project—to examine the health externalities and impact pathways of clean energy development, providing empirical evidence for the social benefits of major infrastructure.

2.2.2. Determinants of Public Health

China has made significant achievements in health construction, continuously improving public health levels and narrowing regional disparities [24]. While existing studies often explore health determinants from a public policy perspective, recent attention has shifted to institutional changes, social security, and individual behaviors. First, regarding public policy, numerous studies employing Difference-in-Differences (DID) methods have found positive health effects from policies such as the “Two Control Zones” environmental regulation [25], the “Air Pollution Prevention and Control Action Plan” [26], and agricultural tax reform [27]. Second, regarding institutional change, research indicates that trade liberalization (e.g., WTO accession) may increase working hours and the probability of illness [28], while changes in medical insurance reimbursement schemes affect patient costs [29]. Third, regarding social security, the literature suggests that improvements in community elderly care [30], pension insurance systems [31], and the structure of health expenditure [32] can enhance resident health, whereas air pollution accelerates the depreciation of health capital [33]. Finally, individual behaviors, such as alcohol consumption [34] and self-employment choices [35], also significantly impact health outcomes. Despite the richness of research on health determinants, the perspective of infrastructure construction remains under-explored, a gap this paper aims to fill using the WEGT as an entry point.

2.2.3. Micro-Level Health Effects of Energy Infrastructure

Energy infrastructure constitutes the physical foundation for energy accessibility, directly determining the pollution exposure and health outcomes of local residents. A growing body of literature has examined the micro-level health impacts of specific energy facilities, generally concluding that non-clean fossil energy infrastructure has detrimental effects on physical health.

For instance, the expansion of coal-fired power plants has been found to lower student health levels, whereas the substitution of clean energy (such as natural gas) improves academic performance [36]. However, some literature notes that natural gas flaring during oil extraction can have a negative impact on health. Studies closely related to this paper include findings from Colombia, where reduced rainfall led to hydroelectric shortages and increased reliance on thermal power, causing air pollution that raised respiratory and cardiovascular morbidity [37]. Similarly, the expansion of natural gas infrastructure in Turkey was found to reduce infant mortality [38]. In India, proximity to coal-fired power plants was associated with respiratory damage [39], while promoting clean cooking fuels improved women’s health [40].

While substantial research exists, much of it focuses on developed contexts or specific developing economies. China, however, faces more severe challenges regarding industrial pollution and urbanization, with higher degrees of environmental pollution and governance complexity. More importantly, few studies have evaluated the micro-health externalities of mega-scale cross-regional infrastructure projects like the WEGT. Whether the existing conclusions are applicable to China’s reality requires further exploration. In particular, as the world’s largest energy consumer and most populous nation, the development of clean energy infrastructure in China holds significant importance for global pollution reduction and the improvement of public health standards.

3. Materials and Methods

3.1. Data Source and Processing

This study utilizes individual-level data from the China Health and Nutrition Survey (CHNS), a longitudinal survey project conducted collaboratively by the Chinese Center for Disease Control and Prevention (CCDC) and the University of North Carolina at Chapel Hill (UNC). The survey covers 15 provinces and municipalities, collecting data across individual, household, and community dimensions [41]. Unlike other micro-surveys, the CHNS provides a unique advantage for this study: it includes not only detailed demographic characteristics (e.g., age, gender, income) but also comprehensive health records, including self-reported illness in the past four weeks and specific chronic conditions. The reliability and objectivity of its health metrics have led to its widespread application in health economics research [42].

For the empirical analysis, we selected four waves of survey data: 2006, 2009, 2011, and 2015. The rationale for this selection is twofold: first, to establish a clean baseline period (2006 and 2009) prior to the full-scale operation of the WEGT Line 2, minimizing interference from antecedent energy policies; and second, to utilize the latest available public data (2015) to capture the post-treatment effects.

We merged the CHNS data with the geographical route information of the WEGT Line 2. Individuals residing in cities traversed by the Pipeline’s main or branch lines were identified as the treatment group, while those in non-traversed cities served as the control group. Additionally, city-level characteristic variables were matched to the dataset using the China City Statistical Yearbook and the China Economic and Social Big Data Platform. The final integrated dataset comprises a balanced panel containing information on project-node and non-node cities across the four selected waves.

3.2. Identification Strategy

To evaluate the causal impact of the WEGT Line 2 on public health, following the approach of Imelda (2020) [43], we treat the operation of the WEGT Line 2 as a quasi-natural experiment. Since the Pipeline’s route planning is primarily determined by national energy strategy and geological conditions rather than local resident health levels, the project serves as an exogenous shock. Therefore, we employ a Difference-in-Differences (DID) model for identification, specified as follows:

$$Y_{ict}=\alpha +\beta WEG{T}_{-Pos{t}_{ct}}+\gamma Contro{l}_{ict}+{\delta }_c+{\tau }_t+{\epsilon }_{ict}$$

(1)

In Equation (1), the subscripts $i$, $c,$ and $t$ denote the individual, city, and year, respectively. The dependent variable, $Y_{ict}$, represents health status, assigned a value of 1 if the individual was sick or injured in the past four weeks, and zero otherwise.

The core explanatory variable is the interaction term $WEG{T}_{-Pos{t}_{ct}}$. Here, $WEGT$ is a dummy variable indicating whether a city is a node on the $WEGT$ Line 2 route (1 for treated cities, 0 for control cities), and $post$ indicates the time period, taking a value of 1 for years after the project’s operation and zero otherwise.

$Contro{l}_{ict}$ Represents a set of control variables covering individual, household, and regional characteristics. ${\delta }_c$ and ${\tau }_t$ Denote city fixed effects and year fixed effects, respectively, to absorb time-invariant regional heterogeneity and time-varying common shocks ${\epsilon }_{ict}$. This is the random error term. Standard errors are clustered at the city level. Our sample comprises 46 distinct city clusters, consisting of 18 treated (node) cities and 28 control (non-node) cities. We cluster at the city level rather than the provincial level for two reasons: first, the treatment assignment (pipeline node status) is determined at the city level; second, the limited Number of provinces (15) in the CHNS sample could lead to downward-biased standard errors if clustered at the provincial level (Cameron et al., 2008) [44]. The coefficient of interest is $\beta$, which is expected to be negative, implying that the $WEGT$ project reduces the illness rate of residents in regions along the Pipeline.

3.3. Variable Selection

3.3.1. Dependent Variable

Extensive literature has established that air pollution is associated with an increased risk of respiratory and cardiovascular diseases [45]. Consequently, objective morbidity indicators for these specific conditions are often used as dependent variables to assess pollution-induced health issues [46]. While respiratory and cardiovascular diseases are direct indicators of air pollution exposure and are widely used in international health studies [47], the CHNS dataset contains a high rate of missing values for these specific disease categories. To address this, following the approach of recent domestic studies on pollution and health [48], we adopt “whether the respondent was sick or injured in the past four weeks” as the primary proxy for resident health in our baseline regression. This self-reported health measure is a widely accepted comprehensive indicator in health economics. We acknowledge that this broad indicator contains measurement error as it encompasses ailments unrelated to air pollution, such as accidental injuries or non-respiratory infections. However, from an econometric perspective, the inclusion of such unrelated health events acts as classical measurement error in the dependent variable. This type of noise typically inflates standard errors and biases the estimated coefficients toward zero (attenuation bias), making it more difficult to detect statistically significant effects. Therefore, the significant health improvements we observe likely represent a conservative lower bound of the true impact of the WEGT project on pollution-sensitive health outcomes.

3.3.2. Key Explanatory Variable

The core independent variable is the interaction term representing whether a city is a node city along the $\mathrm{W}\mathrm{E}\mathrm{G}\mathrm{T}$ Line 2 after the project’s operation. We identified the node cities by cross-referencing the Map of Natural Gas Pipelines in China (2013), published by the Editorial Board of the Yearbook of China’s Gas Industry, with official announcements from municipal governments. This identification strategy relies on objective geographical data to ensure transparency and replicability. To facilitate verification of the treatment assignment, we explicitly list all 46 cities included in our sample, categorized by treatment status, in

Table A7. List of Sample Cities (Treatment vs. Control Group).

Group	Number	Cities
Treatment Group (Node Cities)	18	Pingliang City; Zhangye City; Foshan City; Guangzhou City; Qingyuan City; Shenzhen City; Nanning City; Luoyang City; Pingdingshan City; Huanggang City; Xuzhou City; Jiujiang City; Nanchang City; Shangrao City; Weinan City; Shanghai Municipality; Hangzhou City; Ningbo City
Control Group (Non-Node Cities)	28	Dingxi City; Lanzhou City; Chaozhou City; Jiangmen City; Maoming City; Guilin City; Hechi City; Yulin City; Xinyang City; Zhengzhou City; Zhoukou City; Jingmen City; Yueyang City; Changsha City; Lianyungang City; Suzhou City; Suqian City; Taizhou City; Yancheng City; Yangzhou City; Jingdezhen City; Baoji City; Hanzhong City; Yulin City; Huzhou City; Jiaxing City; Lishui City; Taizhou City

of Appendix A. Based on this, we constructed the interaction term $\mathrm{W}\mathrm{E}\mathrm{G}{\mathrm{T}}_{-\mathrm{P}\mathrm{o}\mathrm{s}{\mathrm{t}}_{\mathrm{c}\mathrm{t}}}$, indicating the treatment status and timing.

We acknowledge that this binary classification does not capture the continuous variation in natural gas penetration rates or secondary intra-provincial distribution. However, we employ this “Intention-to-Treat” (ITT) framework to identify the causal effect of infrastructure availability strictly. Unlike actual gas consumption or penetration rates, which are often endogenous to local economic development and household preferences, the pipeline routing represents an exogenous supply-side shock determined by national strategy. By focusing on the extensive margin (access) rather than the intensive margin (usage volume), we mitigate potential reverse causality. Furthermore, our mechanism analysis in Section 5.1 validates that this binary treatment status serves as a strong predictor for the significant increase in city-level natural gas consumption, ensuring the validity of our identification strategy.

3.3.3. Control Variables

Drawing on existing literature [49], we control for a comprehensive set of covariates at the individual, household, and city levels to mitigate omitted variable bias:

• Individual-level controls: Age, Gender, Years of schooling (Education), Marital status, Occupation type (Agricultural sector), Medical insurance status, and Household registration status (Hukou).
• Household-level controls: Household size and Household income per capita.
• City-level controls: Total population and Number of hospital beds (a proxy for medical resource availability).

3.3.4. Descriptive Statistics Analysis

Table 1. Descriptive Statistics of Variables.

Variable	Definition	N WEGT Nodes (Treatment)	Mean	Std. Dev.	N Non-WEGT Nodes (Control)	Mean	Std. Dev.
Illness Rate	1 if sick/injured in past 4 weeks, zero otherwise	10,909	0.182	0.53	40,537	0.153	0.48
Age	Actual age of the respondent	10,931	44.61	20	40,593	46.48	18.99
Gender	1 for Male, 0 for Female	10,932	0.483	0.5	40,593	0.478	0.5
Education	Years of schooling	10,332	7.901	4.776	38,133	8.306	5.019
Marital Status	1 if Married, zero otherwise	9540	0.886	0.317	36,191	0.924	0.265
Agriculture	1 if agricultural job, zero other jobs	9652	0.196	0.397	36,504	0.204	0.403
Medical Ins.	1 if insured, zero otherwise	10,931	0.799	0.401	40,593	0.856	0.352
Hukou	1 for Urban, 0 for Rural	10,932	0.314	0.464	40,593	0.381	0.486
Income	Household income per capita (log)	10,365	1.887	1.431	38,245	2.042	1.559
Family Size	Number of family members	10,775	4.29	1.816	39,966	3.591	1.533
Population	City population (log/10k)	10,074	1.143	0.673	38,282	1.629	1.686
Hospital Beds	Number of beds (log/10k)	10,717	1.395	1.081	38,760	3.581	4.125

Note: The final analytical sample covers 46 cities, comprising node cities (Treatment) and non-node cities (Control).

presents the descriptive statistics for the main variables.

Table 1 presents the descriptive statistics for the main variables. On average, the illness rate in the control group is slightly lower than that in the treatment group. While demographic characteristics such as age, gender, education, marital status, and household size are comparable between the two groups, notable disparities exist in other dimensions. Specifically, the treatment group exhibits lower levels of medical insurance coverage, urbanization (Hukou), household income, city population, and medical resources (hospital beds) compared to the control group. These baseline disparities suggest potential selection bias, underscoring the necessity of employing the Difference-in-Differences (DID) model with fixed effects to account for time-invariant unobserved heterogeneity and ensure robust causal inference.

4. Empirical Results

4.1. Baseline Estimation Results

Baseline Estimation Results Based on the quasi-natural experiment provided by the full-scale operation of the WEGT Line 2 in 2012, we employ a panel fixed effects model to examine the potential public health benefits of the project. The regression results are reported in

Table 2. Baseline Regression Results.

Variables	(1)	(2)	(3)	(4)
WEGT_Post	−0.033 **	−0.063 **	−0.078 ***	−0.082 ***
	(0.013)	(0.024)	(0.019)	(0.026)
Constant	0.162 ***	0.016	0.166 ***	0.032
	(0.011)	(0.027)	(0.002)	(0.181)
Individual Controls	No	Yes	No	Yes
Household Controls	No	Yes	No	Yes
City Controls	No	Yes	No	Yes
Year FE	No	No	Yes	Yes
City FE	No	No	Yes	Yes
Observations	51,446	40,265	51,446	40,265
Adjusted R2	0.000	0.023	0.013	0.034

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Standard errors clustered at the city level are in parentheses.

. Column (1) presents the estimates without control variables and fixed effects. Columns (2) and (3) introduce control variables and fixed effects separately, while Column (4) incorporates the full set of controls and two-way fixed effects. Across all specifications, the coefficients of the core explanatory variable, $WEG{T}_{-Post}$, are consistently negative and statistically significant at the 5% level or better. This indicates that, compared to non-node cities, the incidence of illness among residents in node cities significantly decreased following the operation of the WEGT Line 2. These results confirm that the increased supply of natural gas brought by the project significantly improved public health in regions along the Pipeline. This finding aligns with the evidence provided by Cesur et al. [50] regarding the health benefits of expanding natural gas infrastructure.

While the aggregate illness rate provides a comprehensive overview of public health status, it is a broad measure that may capture non-environmental health shocks. To provide more direct evidence regarding the pollution-health channel and treat disease-specific outcomes as co-primary results, we further examined two pollution-sensitive morbidity indicators: respiratory diseases and cardiovascular diseases. Extensive medical literature has linked these conditions directly to air quality degradation.

We re-estimated Equation (1) using these specific disease indicators as dependent variables. It is important to note that due to data availability in the CHNS, the sample size for these specific diagnoses is smaller than that for the general illness indicator. The results are presented in

Table 3. Impact of WEGT Line 2 on Specific Disease Outcomes.

Variable	Respiratory Diseases	Cardiovascular Diseases
WEGT_Post	−0.0113 ***	−0.0322 ***
	(0.0028)	(0.0081)
Constant	0.0247	−0.1787 *
	(0.0158)	(0.0921)
Individual Controls	Yes	Yes
Household Controls	Yes	Yes
Regional Controls	Yes	Yes
Year Fixed Effects	Yes	Yes
City Fixed Effects	Yes	Yes
Household Fixed Effects	No	No
Individual Fixed Effects	No	No
Observations	31,000	40,184
Adj. R2	0.0122	0.1306

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

. The coefficients for the core explanatory variable are negative and statistically significant for both respiratory and cardiovascular diseases. Although the reduced sample size warrants a cautious interpretation, these findings provide suggestive evidence that the WEGT Line 2 project contributes to public health improvements, specifically by mitigating conditions closely tied to air quality. This consistency between the broad health measure and specific pollution-related diseases reinforces the plausibility of our baseline estimates.

4.2. Robustness Checks

4.2.1. Parallel Trend Test

The validity of the Difference-in-Differences (DID) method relies on the parallel trend assumption, which requires that, in the absence of the intervention, the treatment and control groups would have followed the same trajectory over time. To verify this, we employed the event study method [51] to examine the dynamic effects of the WEGT Line 2 project. The model is specified as follows:

$$Y_{ict}={\gamma }_0+\sum _k{\theta }_{k}WEGT\times I(t=k)+\phi Contro{l}_{ict}+{\delta }_v+{\tau }_t+{\epsilon }_{ict}$$

(2)

In Equation (2), $I(t=k)$ serves as a dummy variable for the year $k$. Based on the survey waves utilized in this study, $k$ takes the values of 2006, 2009, and 2011 to represent the pre-implementation period (denoted as $WEG{T}_{-Pre3}$, $WEG{T}_{-Pre2}$ and $WEG{T}_{-Pre1}$), and 2015 to represent the post-implementation period (denoted as $WEG{T}_{-Pos{t}_1}$). It is important to note that while the Pipeline became operational in late 2012, we utilize the 2015 wave as the post-treatment observation. This selection accounts for the “implementation lag” required for constructing last-mile distribution networks and the “health response lag” for physical improvements to materialize. Therefore, the 2015 data effectively capture the stable medium-term policy effect following a transition period (2013–2014). The year 2011 ($WEG{T}_{-Pre1}$ is selected as the reference base year. The control variables remain consistent with the baseline regression. The parameter ${\theta }_k$ Captures the difference in health levels between the treatment and control groups relative to the base year.

The results of the parallel trend test are presented in Figure 1. As shown in the figure, the estimated coefficients for the three pre-treatment periods ($WEG{T}_{-Pre3}$, $WEG{T}_{-Pre2}$ and $WEG{T}_{-Pre1}$). fluctuate around zero. Crucially, the 95% confidence intervals for all pre-treatment estimates cross the horizontal zero line, implying that the difference in health outcomes between the treatment and control groups was not statistically significant prior to the policy implementation.

To further rigorously verify the parallel trend assumption and address potential concerns regarding limited statistical power due to the Number of observations, we conducted a joint F-test for the pre-treatment coefficients. The resulting F-statistic is 0.29 (p-value = 0.75), indicating that we cannot reject the null hypothesis that the pre-treatment coefficients are jointly equal to zero.

The intermittent nature of the CHNS waves (2006, 2009, 2011) limits the temporal granularity of the pre-trend analysis. Despite this data limitation, the convergence of evidence—(1) individual coefficients indistinguishable from zero, and (2) the joint F-test failing to reject the null—provides robust support for the parallel trend assumption. In contrast, a significant divergence is observed immediately after the project’s operation ($WEG{T}_{-Pos{t}_1}$), with the coefficient significantly dropping below zero.

4.2.2. Placebo Test

To rule out the possibility that unobservable omitted variables drive the baseline results, we conducted a placebo test by randomly assigning the treatment. Specifically, we randomly selected a set of cities from the sample to form a pseudo-treatment group, while the remaining cities constituted a pseudo-control group. We then re-estimated the baseline model using these generated groups.

This process was repeated 1000 times to obtain a distribution of estimated coefficients and their corresponding p-values. As illustrated in Figure 2, the mean of the randomized coefficients is concentrated around zero and follows a normal distribution. This distribution significantly deviates from the actual estimated coefficient (represented by the vertical dashed line). Furthermore, the majority of the p-values are greater than 0.1. These findings suggest that the baseline results are not driven by random chance or unobserved factors, further validating the robustness of our conclusions.

4.2.3. Addressing Sample Selection Bias

The routing of the WEGT Line 2 project considers factors such as geographical location, resource endowment, and economic development, which means certain cities have a higher probability of becoming project nodes. This non-random placement may introduce sample selection bias. Although our baseline model controls for city-level fixed effects, potential bias may persist. To further mitigate this issue and enhance the exogeneity of the shock, we adopted the following three strategies:

• Excluding Provincial Capitals

Provincial capitals typically possess larger scales and higher levels of economic development, increasing their likelihood of being selected as WEGT Line 2 nodes. Furthermore, these cities generally offer superior medical conditions, which naturally contribute to better public health outcomes. To eliminate this potential bias, we excluded provincial capitals from the sample. The regression results, presented in Appendix A Column (1) of

Table A1. Sample Selection Bias Test Results.

Variable	(1) Excluding Provincial Capitals	(2) Excluding YRD Region	(3) Using Pipeline-Adjacent Provinces Sample
WEGT_Post	−0.0735 ***	−0.0595 ***	−0.0785 ***
	(0.0171)	(0.0202)	(0.0278)
Constant	−0.3957 *	−0.0135	0.5204
	(0.2258)	(0.1728)	(0.3798)
Individual Controls	Yes	Yes	Yes
Household Controls	Yes	Yes	Yes
Regional Controls	Yes	Yes	Yes
Year Fixed Effects	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	Yes
Observations	35,404	33,708	16,944
Adj. R2	0.0317	0.0311	0.0336

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

, show that the coefficient of WEGT_Post remains significantly negative, indicating that the health benefits of the WEGT Line 2 project hold even after removing these special administrative units.

• Excluding the Yangtze River Delta (YRD) Urban Agglomeration

The Yangtze River Delta is one of China’s most economically developed regions, characterized by an industrial structure and energy demand pattern distinct from those of other regions. This uniqueness might mask or amplify the actual impact of the WEGT project. Additionally, the YRD’s rapid economic growth and mature energy infrastructure allow for economies of scale and lower natural gas supply costs, potentially skewing the assessment of health effects. Therefore, we excluded cities in the YRD to remove these confounding factors. The results in Column (1) of Table A1 demonstrate that excluding the YRD sample does not qualitatively change the estimation results.

• Limiting the Sample to Pipeline-Adjacent Provinces

While the natural gas from the WEGT project primarily supplies nodal cities along the main line, it may also be transported to other cities within the same province via provincial pipelines. If non-nodal cities within the same province also benefit from the gas supply, the baseline regression might overestimate the specific health effects of the WEGT Line 2 project (by contaminating the control group). To address this, drawing on the method by [52], we restricted the sample to provinces traversed by the WEGT Line 2 line, excluding non-adjacent provinces. In this subsample, nodal cities along the line served as the treatment group, while other cities within the same provinces served as the control group. The results, reported in Column (3) of Table A1, show that the coefficient of WEGT_Post remains significantly negative. This specific test addresses the concern of ‘partial treatment’ or intra-provincial spillovers. Even if non-node cities within the same province benefit partially from gas redistribution (making them a more competitive control group), the significant treatment effect persists. This confirms that the direct health benefits for node cities substantially outweigh any partial benefits received by neighboring non-node cities.

4.2.4. Excluding Interference from Concurrent Energy Projects

Other major energy infrastructure projects implemented during the sample period could potentially confound the estimated health effects of the WEGT Line 2 project. To mitigate this interference and isolate the net effect of the WEGT Line 2, we sequentially excluded samples affected by other representative natural gas pipelines that became operational around the same time. These primarily include the Shaanxi-Beijing Gas Pipeline III and the Qinhuangdao-Shenyang Gas Pipeline.

• Excluding the Shaanxi-Beijing Gas Pipeline III

The Shaanxi-Beijing Gas Pipeline III commenced construction in May 2009 and began operation in January 2011, running from Yulin City in Shaanxi Province to Liangxiang District in Beijing. Given that its operation time closely coincides with the WEGT Line 2 project, it presents a potential confounding risk. Following the approach of [53], we excluded cities traversed by the Shaanxi-Beijing Gas Pipeline III from the total sample. The regression results, reported in Column (1) of

Table A2. Excluding Interference from Concurrent Energy Projects and Environmental Policies.

Variable	(1) Shaanxi-Beijing Line 3	(2) Qin-Shen Line	(3) Low-Carbon City	(4) Carbon Trading Market	(5) Action Plan (Atmosphere Ten)	(6) Simultaneous Exclusion
WEGT_Post	−0.0997 ***	−0.0803 ***	−0.0706 ***	−0.0652 ***	−0.0606 **	−0.0773 ***
	(0.0244)	(0.0255)	(0.0216)	(0.014)	(0.0241)	(0.0155)
Constant	−0.1674	0.0752	−0.0177	−0.1531 *	−0.0292	−0.1553 *
	(0.1259)	(0.1852)	(0.1249)	(0.0902)	(0.124)	(0.0872)
Individual Controls	Yes	Yes	Yes	Yes	Yes	Yes
Household Controls	Yes	Yes	Yes	Yes	Yes	Yes
Regional Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Observations	38,193	36,623	40,265	40,265	40,265	40,265
Adj. R2	0.0341	0.0359	0.0339	0.0353	0.034	0.0355

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

, show that the estimated coefficient remains consistent with the baseline regression after removing the potential influence of this concurrent project.

• Excluding the Qinhuangdao-Shenyang Gas Pipeline

The Qinhuangdao-Shenyang Pipeline started construction in May 2009 and was put into operation in February 2011. It begins in Qinhuangdao City, Hebei Province, and concludes in Shenyang City, Liaoning Province, with branches extending to Huludao, Panjin, and Shenyang. Similarly, we excluded the sample of cities covered by the main line and branches of this Pipeline to eliminate its interference with the evaluation of the WEGT Line 2’s health effects. The results in Column (2) of Table A2 indicate that the coefficient of WEGT_Post remains significantly negative. This suggests that our baseline conclusion remains robust even after accounting for the impact of the Qinhuangdao-Shenyang natural gas project.

4.2.5. Excluding Interference from Concurrent Environmental Policies

Environmental regulations serve as effective instruments for pollution abatement and public health improvement. During the “11th Five-Year Plan” and “12th Five-Year Plan” periods, China implemented various environmental regulatory policies. The distinct environmental protection measures adopted in pilot regions for these policies could lead to pollution reduction, which might confound the observed health effects in our study. To address this, we selected three representative environmental policies implemented during the sample period—the “Low-carbon City” pilot, the “Carbon Emission Trading Scheme (CETS)” pilot, and the “Air Pollution Prevention and Control Action Plan”—to rule out their potential interference.

• Controlling for the “Low-carbon City” Pilot Policy

China’s Low-Carbon City pilot policy was progressively rolled out to reduce greenhouse gas emissions, improve energy efficiency, and promote sustainable urban planning. Since 2010, 81 cities have been designated as pilot zones, which has significantly improved urban air quality [54]. Following the method of [55], we added an interaction term between the “Low-carbon City” dummy and year dummy variables to the baseline model to control for the policy’s potential impact. The results, shown in Column (3) of Table A2, indicate that the core explanatory variable remains significantly negative, confirming the robustness of our baseline conclusion.

• Controlling for the “Carbon Emission Trading Scheme (CETS)” Pilot Policy

To fulfill its commitment to reducing carbon intensity made at the Copenhagen Climate Change Conference, China began exploring market-based mechanisms to reduce greenhouse gas emissions, with the carbon market being a crucial component. In 2011, the National Development and Reform Commission approved pilot work for carbon emission trading in seven provinces and cities: Beijing, Shanghai, Tianjin, Chongqing, Hubei, Guangdong, and Shenzhen. The market launched progressively starting in 2013. The CETS has generated significant environmental dividends [56] and may also affect resident health. Therefore, we included an interaction term between the CETS pilot dummy variable and the year dummy variable in the regression. The results reported in Column (4) of Table A2 show that the findings remain robust.

• Controlling for the “Air Pollution Prevention and Control Action Plan”

To address increasingly severe air pollution and reduce the concentration of particulate matter, the State Council issued the Air Pollution Prevention and Control Action Plan in September 2013. By January 2018, the goals of the “Action Plan” were fully achieved, effectively reducing air pollutant concentrations and improving public health [57]. To exclude the influence of this policy, we added an interaction term between the dummy variable for key regions targeted by the “Action Plan” and a time dummy (pre- and post-2013) to the baseline model. Column (5) of Table A2 reports the estimated results, which reveal that the baseline conclusion remains unchanged even after controlling for the “Action Plan.”

Finally, to rigorously exclude the interference of all three environmental policies simultaneously, we included the interaction terms for all three policies in the regression model. The results are reported in Column (6) of Table A2. The coefficient of the core explanatory variable remains significantly negative, demonstrating that our main conclusion remains robust even after accounting for the combined influence of these concurrent environmental regulations.

4.2.6. Additional Robustness Checks

To further verify the reliability of our findings, we conducted a series of additional sensitivity analyses, including excluding confounding project impacts, controlling for other energy infrastructure, using alternative dependent variables, and employing high-dimensional fixed effects. The results of these tests are summarized in

Table A3. Results After Excluding Interference from Engineering Projects, Other Energy, and Energy Facility Construction.

Variable	(1) Excluding WEGT Line I	(2) Project Overlapping Areas	(3) Excluding LPG	(4) Excluding Energy Infrastructure	(5) Simultaneous Exclusion
WEGT_Post	−0.0890 ***	−0.0884 **	−0.0928 ***	−0.0817 **	−0.1003 ***
	(0.0277)	(0.0345)	(0.019)	(0.0303)	(0.0219)
Constant	−0.0266	0.3143	−0.0557	−0.2602	−0.1115
	(0.1857)	(0.5045)	(0.1191)	(0.2614)	(0.3642)
Individual Controls	Yes	Yes	Yes	Yes	Yes
Household Controls	Yes	Yes	Yes	Yes	Yes
Regional Controls	Yes	Yes	Yes	Yes	Yes
Year Fixed Effects	Yes	Yes	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	Yes	Yes	Yes
Observations	40,265	33,920	37,473	25,286	24,049
Adj. R2	0.0336	0.035	0.0368	0.0475	0.0495

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Standard errors clustered at the city level are in parentheses.

and

Table A4. Results for Alternative Dependent Variables, Excluding Omitted Variables, and Changing Fixed Effects.

Variable	City × Time	Province × Time	Household FE	Individual FE
WEGT_Post	−0.0697 ***	−0.0906 ***	−0.0883 ***	−0.0868 *
	(0.0205)	(0.0217)	(0.0272)	(0.0319)
Constant	−0.0513	0.06517	0.03376	0.0137
	(0.1299)	(0.1306)	(0.0539)	(0.0422)
Individual Controls	Yes	Yes	Yes	Yes
Household Controls	Yes	Yes	Yes	Yes
Regional Controls	Yes	Yes	Yes	Yes
Year Fixed Effects	Yes	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	No	No
Household Fixed Effects	No	No	Yes	No
Individual Fixed Effects	No	No	No	Yes
Observations	40,265	40,265	39,530	30,792
Adj. R2	0.0342	0.034	0.0527	0.0294

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Standard errors clustered at the city level are in parentheses.

.

• Excluding the Impact of WEGT I

The First West–East Gas Pipeline (WEGT I) was completed and put into operation in 2004. Although our sample period begins in 2006 to minimize its immediate impact, there is a geographical overlap between the two phases. To eliminate the potential interference of the first phase, following the method of [58], we excluded cities traversed by WEGT I and re-estimated the model. The results remain robust after removing these samples.

• Analysis of Overlapping Areas

To investigate whether the health effects in areas covered by both WEGT I and WEGT Line 2 differ from the baseline, we conducted a subsample analysis. We defined cities traversed by both pipelines as the treatment group and cities not covered by either Pipeline as the control group, excluding those covered only by WEGT I or only by WEGT Line 2. The estimated coefficient remains significantly negative and is similar in magnitude to the baseline result, indicating that the existence of WEGT I does not bias the main conclusions regarding WEGT Line 2.

• Controlling for Other Clean Energy and Infrastructure

The consumption of other clean energy sources might substitute for natural gas, and local energy infrastructure could impact natural gas penetration rates. To account for this, we obtained data on the total supply of liquefied petroleum gas (LPG) and natural gas storage capacity from the China City Statistical Yearbook and China Urban Construction Statistical Yearbook. These variables were added as controls in the baseline model. The results indicate that the conclusion remains robust after controlling for these alternative energy sources and infrastructure factors.

• Mitigating Omitted Variable Bias

Although the baseline model controls for multidimensional characteristics, public health may still be influenced by unobserved seasonal factors or time trends. To mitigate bias from time-correlated omitted variables, drawing on [59], we added city-specific linear time trends and province-by-year fixed effects to the model. These controls account for distinct economic growth paths across cities and province-specific shocks over time. The coefficient of the core explanatory variable remains significantly negative, suggesting that omitted variable bias does not drive our main results.

• Alternative Fixed Effects

To more effectively control for unobserved time-invariant heterogeneity at the micro-level, we replaced the city fixed effects in the baseline model with household fixed effects and individual fixed effects. This specification controls for factors such as genetics, long-term family habits, and other persistent individual characteristics. The results demonstrate that, even with these stringent fixed effects, the coefficient of the policy variable remains significantly negative, further confirming the robustness of the estimates.

• Wild Cluster Bootstrap

Although our total number of clusters (46) exceeds the conventional rule-of-thumb threshold of 42 suggested by [60] for reliable inference, the number of treated clusters (18) is relatively limited. To rigorously address potential small-sample bias and verify the reliability of our inference, we employed the Wild Cluster Bootstrap method with 1000 replications. The test yielded a p-value of 0.001 for the core explanatory variable (WEGT_Post), which remains statistically significant. This confirms that our main findings are robust and not driven by the finite number of clusters.

5. Mechanism Analysis and Heterogeneity Discussion

5.1. Mechanism Analysis

The preceding analysis established that the WEGT Line 2 significantly improved the health levels of residents in nodal cities. Given that the causal link between air pollution and morbidity is well-established in the epidemiological literature [61], demonstrating that the WEGT project causally mitigates pollution sources would provide a plausible logical mechanism for the observed health improvements.

To elucidate the underlying pathways driving this result, we adopt a ‘chain of evidence’ approach to investigate the intermediate determinants of health, utilizing a systematic framework that spans both micro and macro dimensions. Methodologically, it is crucial to note that we estimate the effect of the policy on these intermediate variables separately, rather than including them as controls in the baseline model. This strategy allows us to identify the policy’s transmission channels while strictly avoiding collider bias.

Structurally, our analysis proceeds in three logical steps to cross-validate the pollution-reduction mechanism. First, at the micro-level, we examine whether the project facilitates fuel switching and energy structure optimization within households (Section 5.1.1). Second, we investigate whether industrial firms in node cities reduce their emission intensity of key pollutants due to access to cleaner energy (Section 5.1.2). Finally, we corroborate these micro-level behavioral changes by verifying aggregate environmental improvements at the macro-level of cities (Section 5.1.3). This multidimensional approach ensures that the health externalities are supported by consistent evidence from individual energy choices, corporate emissions, and regional air quality.

5.1.1. Household Energy Choice

Indoor air pollution is a leading risk factor for public health in many developing countries [62]. Household energy consumption constitutes a significant portion of China’s energy demand [63], and an irrational energy structure directly affects resident health. This is primarily because the combustion of solid fuels, such as coal and biomass, releases substantial amounts of indoor pollutants that are easily inhaled [64]. Therefore, promoting a transition toward clean household energy is a practical pathway to improve family health.

To verify this channel, we matched the full sample with household energy usage data. Following the classification standards of [65] and the World Health Organization’s guidelines on indoor air quality [66], we categorized household fuels as follows: “Traditional Energy” (charcoal, firewood, and crop residue/straw), “Polluting Energy” (unprocessed solid fuels like coal, and kerosene), and “Clean Energy” (Natural Gas). We constructed binary variables for the usage of these energy types and estimated the effects using the baseline model, controlling for household fixed effects to absorb time-invariant heterogeneity.

The regression results are reported in

Table 4. Household Energy Choices.

Variable	(1) Traditional Energy	(2) Polluting Energy	(3) Natural Gas
WEGT_Post	−0.028 **	−0.046 ***	0.062 ***
	(0.013)	(0.012)	(0.011)
Constant	0.363 ***	0.783 ***	0.183 ***
	(0.035)	(0.027)	(0.019)
Individual Controls	Yes	Yes	Yes
Household Controls	Yes	Yes	Yes
Regional Controls	Yes	Yes	Yes
Household FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Observations	14,154	14,154	14,154
Adj. R2	0.499	0.562	0.639

Note: Standard errors clustered at the city level are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. “Traditional Energy” includes charcoal, firewood, and crop residue; “Polluting Energy” includes coal and kerosene. Standard errors clustered at the city level are in parentheses.

. Compared to households in non-nodal cities, the WEGT Line 2 project significantly reduces the probability of using traditional and polluting energy in nodal cities, while significantly increasing the probability of using natural gas (statistically significant at least at the 5% level). This result indicates that the WEGT Line 2 project has effectively promoted the upgrading of the household energy structure. Since substituting dirty fuels with clean fuels reduces the incidence of illness among family members [67], our findings suggest that facilitating a clean energy transition at the household level is a plausible channel through which the project improves public health.

5.1.2. Firm Pollution Reduction

Industrial energy consumption is a primary source of pollution emissions. Promoting the transition toward cleaner energy structures in the industrial sector is a crucial pathway to reconciling production growth with environmental protection. The WEGT Line 2 project provides an abundant supply of clean energy to firms in nodal cities, facilitating the clean transformation of their energy consumption structure [68] and reducing pollution emissions, thereby improving local public health.

To explore this pollution-reduction mechanism at the firm level, we matched the Annual Survey of Industrial Firms (ASIF) with the Environmental Statistics Database (ESD) to examine the impact of the WEGT Line 2 project on corporate emissions. Following the approach of [69], we selected five pollution indicators: sulfur dioxide (SO₂), chemical oxygen demand (COD), smoke and dust, industrial wastewater (H₂O), and nitrogen oxides (NOX) to capture the impact on various pollutant types.

The estimation results in

Table 5. Firm Pollution Reduction.

Variable	(1) SO2	(2) COD	(3) Smoke	(4) H2O	(5) NOX
WEGT_Post	−0.252 **	−0.337 ***	−0.197 *	−0.740 **	−0.185 **
	(0.121)	(0.061)	(0.119)	(0.292)	(0.088)
Constant	92.829 ***	6.191 ***	−6.621 ***	97.291 ***	−15.970 ***
	(6.519)	(0.893)	(0.321)	(14.547)	(0.71)
Individual Controls	No	No	No	No	No
Household Controls	No	No	No	No	No
Regional Controls	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes
Industry FE	Yes	Yes	Yes	Yes	Yes
Observations	269,655	247,937	270,263	307,449	255,676
Adj. R2	0.246	0.085	0.069	0.051	0.278

Note: Standard errors clustered at the city level are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

show that the WEGT Line 2 project significantly reduces the emissions of SO₂, COD, smoke and dust, wastewater, and NOX. The coefficients of WEGT_Post are statistically significant. This implies that firms in regions traversed by the WEGT Line 2 project are better able to utilize clean energy, thereby reducing pollution emissions and improving public health.

5.1.3. City Environmental Performance

With the advancement of the “Healthy China” strategy and the construction of healthy cities, the importance of urban environmental performance for public health continues to rise. Among environmental performance indicators, urban air pollutant emissions are a direct cause of many public health issues, and these emissions depend on the environmental behaviors of micro-entities, such as firms and households. The WEGT Line 2 project increases the share of natural gas in the energy supply of nodal cities, promoting an upgrade in the urban energy consumption structure.

While the previous sections verified the pollution reduction effects at the household and firm levels, investigating the mechanism at the city level helps validate these micro-channels from a macro perspective. Drawing on [69], we selected sulfur dioxide SO₂, smoke and dust, and PM2.5 as the primary proxies for urban air pollutants. Additionally, we used urban natural gas (NG) consumption as a proxy for the transition in energy structure.

The regression results in Columns (1)–(3) of

Table 6. City Environmental Performance.

Variable	(1) SO2	(2) Smoke	(3) PM2.5	(4) NG
WEGT_Post	−3.723 ***	−2.500 ***	−2.152 ***	4.390 ***
	(0.915)	(0.694)	(0.675)	(1.223)
Constant	7.163 *	5.779	46.762 ***	−16.375 ***
	(3.778)	(3.854)	(1.893)	(4.174)
Individual Controls	No	No	No	No
Household Controls	No	No	No	No
Regional Controls	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
City FE	Yes	Yes	Yes	Yes
Observations	3611	3613	3816	3107
Adj. R2	0.612	0.066	0.945	0.840

Note: Standard errors clustered at the city level are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

indicate that the WEGT Line 2 project significantly reduces urban emissions of SO₂, smoke and dust, and PM2.5. Furthermore, the result in Column (4) confirms that the increased natural gas supply brought by the project effectively promotes a cleaner transformation of the urban energy consumption structure. This result is consistent with the hypothesis that the WEGT Line 2 project improves resident health by optimizing the energy mix and reducing pollution emissions at the city level.

5.2. Heterogeneity Analysis

Given the variations in natural gas accessibility across regions and the differing physiological sensitivities to air pollution among demographic groups, the health benefits of the WEGT Line 2 project may not be uniform. Therefore, we examine the heterogeneity of these effects based on urban-rural attributes and age groups.

5.2.1. Urban-Rural Heterogeneity

The primary objective of the WEGT project is to alleviate the imbalance between regional economic development and energy supply. Cities are characterized by high population density and economic agglomeration. Their rapid development and substantial energy demand allow for economies of scale in natural gas supply. In contrast, rural areas are scattered, making the construction of supporting pipeline infrastructure difficult and costly, leading to a historically low share of natural gas consumption.

To investigate this disparity, we divided the full sample into two subsamples: urban and rural. The regression results, presented in

Table A5. Urban-Rural Attributes.

Variable	Rural		Urban
	(1)	(2)	(3)	(4)
WEGT_Post	−0.0716 ***	−0.0657 ***	−0.0861 ***	−0.0989 ***
	(0.0194)	(0.0189)	(0.0279)	(0.0342)
Constant	0.1581 ***	−0.1652	0.1795 ***	0.1573
	(0.0016)	(0.1653)	(0.0032)	(0.2257)
Individual Controls	No	Yes	No	Yes
Household Controls	No	Yes	No	Yes
Regional Controls	No	Yes	No	Yes
Year Fixed Effects	Yes	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	Yes	Yes
Observations	32,577	24,887	18,869	15,378
Adj. R2	0.0135	0.0312	0.0144	0.0402

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Standard errors clustered at the city level are in parentheses.

, show that the coefficients of the interaction term WEGT_Post are significantly negative for both subsamples, indicating that the WEGT Line 2 project improves health outcomes in both urban and rural areas. However, the health benefits are more pronounced in urban areas compared to rural ones. We further conducted a coefficient difference test, which confirmed that this gap is statistically significant. The results suggest that while cities benefit most directly, the improvement in regional air quality driven by clean energy development in cities also generates positive spillover effects for adjacent rural residents.

5.2.2. Age Heterogeneity

Different age groups exhibit distinct sensitivities to environmental pollution. Theoretically, older adults often have weaker immune systems and pre-existing conditions, making them more susceptible to the adverse effects of air pollution [70]. To identify the differential impacts of the WEGT Line 2 project across age demographics, we examined the heterogeneity from the perspective of age structure.

Drawing on [71], we categorized the sample into three subgroups: children (1–15 years old), adults (16–59 years old), and the elderly (60 years and above). The regression results in

Table A6. Age Differences.

Variable	Children (1–15 Years)		Adults (16–59 Years)		Elderly (60+ Years)
	(1)	(2)	(3)	(4)	(5)	(6)
WEGT_Post	−0.0964 ***	−0.0824 *	−0.0533 ***	−0.0646 ***	−0.1383 ***	−0.1351 ***
	(0.0321)	(0.046)	(0.0185)	(0.0223)	(0.0402)	(0.0498)
Constant	0.1086 ***	0.6587 **	0.1326 ***	−0.0298	0.2714 ***	0.1857
	(0.0032)	(0.2917)	(0.0015)	(0.1878)	(0.0046)	(0.2112)
Individual Controls	No	Yes	No	Yes	No	Yes
Household Controls	No	Yes	No	Yes	No	Yes
Regional Controls	No	Yes	No	Yes	No	Yes
Year Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
City Fixed Effects	Yes	Yes	Yes	Yes	Yes	Yes
Observations	4757	3505	33,269	29,238	13,419	11,027
Adj. R2	0.0103	0.0194	0.012	0.0181	0.026	0.0386

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Standard errors clustered at the city level are in parentheses.

reveal that the WEGT Line 2 project generates health benefits across all three groups, regardless of whether control variables are included. However, the magnitude of the health improvement is largest for the elderly group. This finding aligns with the literature suggesting that pollution damage increases with age [72] and corroborates our mechanism, which suggests that the project improves health primarily by reducing air pollution.

This finding has significant policy implications. As China faces the challenge of rapid population aging, maintaining and improving the health of the elderly is critical. Our results suggest that promoting clean energy development is an effective pathway to achieving “Healthy Aging” and mitigating the health risks associated with demographic shifts.

6. Further Analysis

6.1. Synergistic Effect of the “Coal-to-Gas” Policy

Achieving a green and low-carbon transformation of the energy structure requires simultaneous efforts on both the “energy supply” and “energy consumption” sides. While the WEGT Line 2 project ensures a sufficient supply of clean energy for nodal cities, its pollution reduction effect ultimately depends on the actual utilization of natural gas. The “Coal-to-Gas” (CTG) switching policy, through administrative promotion and subsidies, incentivizes the substitution of coal with natural gas, thereby promoting local clean energy development and reducing pollution emissions [73].

In practice, local governments have not issued a uniform national policy document for the CTG initiative, and implementation varies by region. To examine the synergistic effect of the $\mathrm{C}\mathrm{T}\mathrm{G}$ policy on the health benefits of the WEGT project, drawing on the method of [74], we constructed a dummy variable, $\mathrm{C}\mathrm{T}\mathrm{G}$. This variable takes the value of 1 if a region experienced a decrease in coal consumption and an increase in natural gas consumption concurrently, and zero otherwise. We then established the following Triple Difference (DDD) model:

$$\begin{array}{cc}{Y}_{ict}& =\alpha +{\beta }_{1}WEG{T\_Post\_CTG}_{ct}+{\beta }_{2}WEG{T\_Post}_{ct}+{\beta }_{3}WEG{T\_CTG}_{ct}\\ & + {\beta }_{4}Post{\_CTG}_{ct}+\gamma Contro{l}_{ict}+{\delta }_c+{\tau }_t+{\epsilon }_{it}\hfill \end{array}$$

(3)

In Equation (3), aside from the inclusion of the $CTG$ policy variable, all other variables remain consistent with the baseline regression. The core explanatory variable is the triple interaction term $WEGT\_Post\_CTG$.

The regression results are reported in

Table 7. Synergistic Effect of the “Coal-to-Gas” Policy.

Variable	(1)	(2)	(3)	(4)
WEGT_Post_CTG	−0.098 ***	−0.115 ***	−0.111 ***	−0.136 ***
	(0.000)	(0.004)	(0.006)	(0.014)
Constant	0.145 ***	−0.001	0.162 ***	0.020
	(0.011)	(0.028)	(0.001)	(0.170)
Individual Controls	No	Yes	No	Yes
Household Controls	No	Yes	No	Yes
Regional Controls	No	Yes	No	Yes
Year FE	Yes	No	Yes	Yes
City FE	Yes	No	Yes	Yes
Observations	51,446	40,265	51,446	40,265
Adj. R2	0.003	0.026	0.013	0.034

Note: Standard errors clustered at the city level are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

. Column (1) presents the results without control variables and fixed effects, while Columns (2) and (3) sequentially add controls and fixed effects. Column (4) includes the full set of controls and fixed effects. The estimates show that the absolute value of the coefficient for $WEGT\_Post\_CTG$ is larger than that of the baseline $WEGT\_Post$ coefficient, and it remains statistically significant. This indicates that the CTG policy exerts a synergistic effect (or “amplifying effect”), suggesting that combining energy infrastructure construction with demand-side policy incentives can further accelerate the clean energy transition and enhance the health benefits of the WEGT project.

We acknowledge that constructing the CTG variable based on realized consumption changes (‘ex-post’ measure) introduces potential endogeneity. Specifically, the observed fuel switching could be partially endogenous to the WEGT project itself (i.e., the pipeline availability facilitates the switching). Due to the fragmented nature of local CTG policies and the lack of unified public announcement data during the sample period, we were unable to employ ‘policy announcement dates’ as a strictly exogenous instrument. Therefore, the results in Table 7 should be interpreted as suggestive evidence of the synergistic association between infrastructure supply and realized energy substitution, rather than a causal evaluation of the CTG administrative policy itself.

6.2. Economic Co-Benefits Analysis

Illness imposes not only physical suffering but also significant economic burdens, such as medical expenses and productivity losses. Therefore, focusing solely on the health metrics might underestimate the broader value of the WEGT Line 2 project. To provide a more comprehensive assessment of the project’s externalities, we specifically analyze two economic co-benefits associated with the health improvements: medical costs (direct impact) and employment outcomes (indirect impact).

First, regarding the direct economic burden of disease, we examined both individual medical expenses and city-level medical expenditures. The regression results are reported in Columns (1) and (2) of

Table 8. Economic Co-benefits Analysis.

Variable	(1) Individual Medical Expenses	(2) City Medical Expenditure	(3) Employment Probability
WEGT_Post	−5.401 **	−0.257 **	0.069 ***
	(2.408)	(0.130)	(0.017)
Constant	−1.11	−3.457 ***	0.812 ***
	(3.171)	(1.576)	(0.118)
Individual Controls	Yes	No	Yes
Household Controls	Yes	No	Yes
Regional Controls	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
City FE	Yes	Yes	Yes
Observations	41,294	2400	37,759
Adj. R2	0.001	0.959	0.434

Note: Standard errors clustered at the city level are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

. The findings indicate that the WEGT Line 2 project is significantly associated with reductions in both personal and regional medical costs. Specifically, compared to non-nodal cities, residents in nodal cities saved an average of approximately 540 RMB in medical expenses, and the aggregate city-level medical expenditure for all treated cities decreased by approximately 2.57 billion RMB. It is important to interpret the aggregate city-level reduction with caution. While our fixed-effects model controls for time-invariant characteristics, this reduction may reflect a combination of health improvements and other concurrent factors, such as changes in medical insurance accounting standards or hospital expansion dynamics.

Second, regarding the indirect economic impact manifested through labor market outcomes, we used individual employment status as the dependent variable. The results in Column (3) of Table 8 show that the WEGT Line 2 project significantly increases the probability of individual employment, with a coefficient of 0.069 (significant at the 1% level). We acknowledge that this positive employment effect likely represents the net impact of two mechanisms: the “health channel,” where improved physical condition enhances labor supply capabilities, and the “local economic stimulus channel,” where infrastructure investment boosts local labor demand. While our current identification cannot strictly isolate the health-driven effect from the broader economic stimulus, the results confirm that the project generates positive economic externalities for local communities.

Finally, these estimates likely represent a conservative lower bound of the project’s comprehensive economic value. Many hidden costs associated with disease—such as psychological distress (pain and suffering), efficiency losses due to presenteeism, and long-term sequelae—are difficult to observe and quantify. Therefore, the actual economic benefits generated by the WEGT Line 2 project are likely far greater than these quantifiable estimates suggest.

7. Discussion

This study empirically investigates the health effects of clean energy infrastructure development by leveraging the “quasi-natural experiment” of the WEGT Line 2 project. Consistent with previous macro-level studies, we confirm that the energy transition significantly reduces pollution emissions. However, unlike studies that focus solely on environmental metrics, our micro-level evidence from the CHNS data reveals that these environmental improvements translate directly into substantial public health benefits.

Mechanistically, these results validate the ‘environmental health hypothesis’ by establishing a complete chain of evidence, from infrastructure investment to the optimization of household/industrial energy structures, leading to reduced emissions and ultimately resulting in improved individual health. Furthermore, the heterogeneity analysis highlights that the elderly population benefits the most, providing a compelling policy perspective for addressing the health challenges of an aging society.

Alternative Explanations, Conservative Estimation, and Metric Interpretation. Crucially, to ensure the robustness of our causal interpretation, we must consider potential alternative explanations and clarify the nature of our estimates.

First, we acknowledge the possibility of ‘partial treatment,’ where non-node cities might access natural gas through secondary pipelines (spillovers). Econometrically, if the control group partially receives the treatment, the health disparity between the treatment and control groups would narrow, biasing our DID estimates toward zero (attenuation bias). Therefore, the statistically significant health improvements we observe exist despite this potential spillover.

Second, regarding the dependent variable, one might argue that the broad indicator “sick or injured” captures non-pollution-related health shocks (e.g., accidental injuries). However, from an econometric perspective, the inclusion of such unrelated health events introduces random measurement error. Theoretically, this ‘noise’ leads to attenuation bias, which inflates standard errors and pushes the estimated coefficients toward zero. Consequently, the fact that we still detect a statistically significant negative effect implies that the WEGT’s true impact on pollution-sensitive conditions is likely substantial enough to overcome this noise. Combined, these factors suggest that our baseline estimates likely represent a conservative lower bound of the true policy effect.

Beyond econometric robustness, it is equally important to properly interpret the scope of these health metrics. While our analysis of specific disease indicators (Table 3) verifies the pathological link between air quality and respiratory/cardiovascular conditions, the primary measure used in this study—self-reported general health—captures a broader spectrum of health welfare than clinical diagnoses alone. Distinct from administrative hospital records that track acute medical interventions, self-reported health reflects the overall physical sensation and quality of life, encompassing sub-health symptoms (e.g., discomfort, fatigue) caused by pollution that may not trigger a formal hospital visit. Thus, our findings should be interpreted as an improvement in the “comprehensive general health status” of the population, serving as a valid proxy for environmental welfare.

Combined, these two factors—potential spillovers and measurement error—suggest that our baseline estimates likely represent a conservative lower bound of the true policy effect, rather than an overestimation.

Limitations and Future Research. Despite the robust findings, this study has limitations. First, due to data availability in the CHNS database, our sample period ends in 2015. Consequently, our post-treatment identification relies on a single wave (2015). While this captures the medium-term effect following the implementation lag, this ‘snapshot’ nature limits our ability to assess the long-term stability of the health benefits or fully rule out specific shocks concurrent with that year.

Second, our treatment definition relies on pipeline routing (availability) rather than actual household consumption volume (exposure intensity). While employing this binary ‘Intention-to-Treat’ approach is crucial for mitigating endogeneity—as national strategy rather than local preferences determine pipeline routing—it captures only the extensive margin of the policy effect. Consequently, our estimates reflect the average impact of infrastructure access rather than a precise dose–response relationship based on usage intensity.

Third, constrained by the confidentiality of respondent geocoordinates, we were unable to link individual health outcomes to localized air quality monitors. Instead, following the methodological approach of recent comparable studies, we relied on city-level environmental performance and household-level fuel switching data to validate the mechanism.

Fourth, although we argue that self-reported health captures broad welfare effects, it lacks clinical precision. While we supplemented this with specific respiratory and cardiovascular outcomes, the sample size for these specific diagnoses was limited. Future research utilizing administrative hospital records or medical insurance claims data would offer higher precision. Finally, while we controlled for major concurrent policies, unobserved time-varying local shocks may still exist. Future research could encompass more recent energy infrastructure projects to validate and extend these findings.

8. Conclusions and Policy Implications

8.1. Conclusions

Clean energy development and public health improvement are not only key goals of the UN 2030 Agenda for Sustainable Development but also intrinsic requirements for China’s energy revolution. Based on the integration of WEGT Line 2 nodal city data and CHNS micro-survey data, this paper provides empirical evidence for the synergistic realization of the United Nations Sustainable Development Goals (SDGs). Specifically, our findings illustrate a clear logical pathway where investing in resilient infrastructure (SDG 9) serves as the prerequisite foundation that enables affordable clean energy (SDG 7), which ultimately delivers substantial dividends for good health and well-being (SDG 3). The specific conclusions are as follows:

Significant Improvement in General Health Status: The operation of the WEGT Line 2 project significantly reduces the probability of self-reported illness among residents in nodal cities compared to non-nodal areas.

Dual Mechanisms: The project improves health by promoting the transition to clean household fuels (reducing indoor pollution) and providing clean energy for firms (reducing industrial emissions).

Heterogeneous Benefits: The health dividends are more pronounced in urban areas and among the elderly population, who are more susceptible to air quality issues.

Policy Synergy and Welfare: The “Coal-to-Gas” policy amplifies the health benefits of the infrastructure. Furthermore, the project generates significant social welfare by reducing individual medical expenses by approximately 540 RMB and city-level medical expenditures by 2.57 billion RMB, while also boosting local employment.

8.2. Policy Implications

Based on the empirical findings of this study, we propose three targeted policy recommendations to maximize the health externalities of clean energy infrastructure.

First, policy priority must shift from trunk line construction to the “last mile” of terminal access capability. Our mechanism analysis confirms that the health benefits of the WEGT largely stem from the micro-level substitution of household fuels and the reduction in industrial emissions. Therefore, ensuring that infrastructure effectively reaches end-users is paramount. For households, the government should consider subsidizing pipeline connection fees and the purchase of gas appliances for low-income families, thereby removing the financial barriers to fuel switching documented in our energy choice model. Similarly, for the industrial sector, specific tax incentives or green credit channels should be established to encourage firms to retrofit coal-fired boilers to natural gas, ensuring that the potential emission reductions verified in our firm-level analysis are fully realized. Furthermore, the physical modernization of energy equipment should be synergized with digital integration. Policies should incentivize the deployment of smart meters and IoT sensors to capture high-frequency data on usage and emissions, thereby enhancing the accuracy of data collection. As emerging scholarship underscores, integrating these real-time monitoring tools with Artificial Intelligence (AI) fundamentally revolutionizes cleaner production, optimizing energy consumption patterns to maximize the environmental efficiency of the infrastructure [75].

Second, a “Hard + Soft” synergistic strategy should be implemented to amplify health dividends. The results from our Triple Difference (DDD) model indicate that the “Coal-to-Gas” (CTG) policy has a significant impact on the health effects of the pipeline infrastructure. This implies that infrastructure investment (“Hard Power”) alone is insufficient without complementary regulatory policies (“Soft Power”). Consequently, local governments should synchronize the rollout of pipeline networks with strict environmental regulations, such as the designation of coal-combustion restricted zones. Infrastructure construction plans should be bundled with regional coal-cap targets to ensure that the newly available gas supply effectively displaces coal rather than merely adding to total energy consumption.

Furthermore, a holistic policy framework must encompass cross-sectoral emission reduction strategies. Recent evidence from infrastructure engineering suggests that optimizing carbon emissions throughout the entire project lifecycle, specifically in construction processes and materials, is crucial [76]. Therefore, green construction standards should be applied to the expansion of energy networks to ensure that the building process itself aligns with low-carbon sustainability goals.

Finally, energy transition policies should be integrated with “Healthy Aging” strategies and regional equity goals. Our heterogeneity analysis reveals that the elderly population reaps the largest health dividends from the project. As China faces rapid population aging, policymakers should view clean energy as a critical public health tool and incorporate gas retrofitting into “Old Urban Community Renovation” programs, prioritizing access for nursing homes and communities with high elderly dependency ratios. Furthermore, to address the observed disparity in health benefits between urban and rural areas, future infrastructure phases must bridge the energy divide. A “Universal Service Fund” could be established to cross-subsidize pipeline construction in less densely populated rural regions, ensuring that the health dividends of clean energy are shared equitably across all demographics.